Solvation time of the electron in polar liquids ... - ACS Publications

The formation of esor from its precursor, et~, has been investigated in the temperature range from —65 to. 60°C using a stroboscopic pulse radiolys...
0 downloads 0 Views 1MB Size
Solvation Time of the Electron in Polar Liquids

2835

Solvation Time of the Electron in Polar Liquids. Water and Alcohols W. John Chase’ and John W. Hunt+ The Ontario Cancer Institute and Department of Medical Blophysics, University of Toronto, Toronto, Ontario, M4X 7K9 Canada (Received August 25, 1975) Publication costs assisted by The Ontario Cancer Institute

The formation of esol- from its precursor, et-, has been investigated in the temperature range from -65 to 6OoC using a stroboscopic pulse radiolysis (SPR) system. The rate of formation of esol-, measured at 600 nm, and rate of decay of et-, measured at 1300 nm, are the same; the rate of decay of et- is slower by as much as a factor of 2 when measured at 1050 nm. A t 2OoC the values of ~,,1 are 10.7 f 1,23 f 2,34 f 3, and 39 f 5 psec for methanol, ethanol, 1-propanol, and 1-butanol, respectively. At 2OoC ~~~1 is estimated to be less than 3 psec for H2O and D20. In all of the alcohols tested there is a good correlation between T,,I and the dielectric relaxation time, 72, for rotation of the molecules over a wide range of temperatures; this suggests that intermolecular hydrogen bonding does not limit the rate of electron solvation.

Introduction gen bonding should affect the rate of molecular reorientation. The variation of the solvation rate with temperature Electrons can be separated from their parent molecules should indicate whether hydrogen bonding hinders the moby radiation or chemical techniques. These electrons will lecular reorientation process. interact with solvent molecules and, in polar solvents, most Several laboratories have made attempts to obtain the of them will become trapped in a “cage” of solvent moleformation times of esol- in water and alcohols.’ 1~13,23-27 cules to form a relatively stable species, the solvated elecUsing their stroboscopic pulse radiolysis system, Bronskill tron, esol-.2 In polar solvents such as H20, alcohols, and et al.23 observed the usual absorption spectrum of esol-, ammonia, esol- is a well-established entity on theoretical and estimated that it was formed in less than 10 psec in and experimental grounds.3~~ There is also direct evidence water and alcohols. Beck and Thomasz4 used a fast pulse that a precursor of esol- called the “trapped” or “damp” radiolysis system to obtain the formation of esol- in alcoelectron,2 et-, has been observed in cold glasses and liqhols; values of 2-5 and 50 psec were obtained for ethanol u i d ~ . ~As- ~well, ~ indirect evidence exists from electron and 1-propanol. Recently, Kenney-Wallace and Jonah25 scavenging results that precursors of esol- take part in showed that in normal aliphatic alcohols, the formation early electron proce~ses.~5-~9 time of esol- increases with increasing chain length. BaxHigashimura and his coworkers studied y-irradiated alendale and Wardmanll monitored the formation of esolcohol glasses at 4 K and found a strong absorption band indirectly by observing the concurrent decay of et- at 1300 which peaked at infrared wavelengths.6-s This band shifts nm in cold, liquid alcohols. Gilles et al.13 observed the forirreversibly toward blue wavelengths upon warming to 77 mation of esol- in cool alcohols, and using the data of BaxK. A t the same time, the ESR line width increases, indicatendale and Wardmanll extrapolated to find the formation ing that the cavity around the electron decreases in size times of esol- at 294 K. In water, Kenney-Wallace and and that the electron is stabilized in a deeper trap.6-s Pulse Walker26estimated that the formation time of eaq- was less radiolysis of alcoholic glasses and liquids shows that the than 5.5 psec. Rentzepis et al.27 used a unique picosecond disappearance of the infrared absorption is accompanied mode-locked laser system where the electrons were proconcurrently by the appearance of the absorption band of duced by the photoionization of a solute molecule. Their esol-.9-14 In addition to these experimental observations, studies indicate that esol- is formed in water in about 2 theoretical calculations also support the concept of a parpsec. tially stabilized electron, et-. In ethanol, Fueki et a1.20preThere were considerable discrepancies among the values dicted an optical absorption shift due to dipole orientation for the formation times of esol-: for example, values from of solvent molecules around the electron; this shift was at 2-524 to 18 psec26have been published for ethanol. Acculeast 70% of the observed value. Tachiya et aL21 have rate measurements of the formation times are needed to shown that excess electrons in a polar glass would initially elucidate the mechanism of the electron solvation process. be trapped in a relatively shallow preexisting trap, a particIn particular, we have studied the formation of esol- in ular arrangement of the molecular dipoles forming a region H20, DzO, ethylene glycol, and the first four normal alcoof low potential energy for the electron. This model indihols by observing the absorption signals from our pulse racates that virtually all of the trapping sites would have diolysis system. energies less than 1 eV, well below the energy of the peak absorption of esol- (1.77 eV in ethanol22). Photobleaching pulse radiolysis s t ~ d i e s , ~ - l ~Experimental Section and theoretical calculations20v21indicate that molecular oriThe stroboscopic pulse radiolysis (SPR) system allows us entation, rather than thermal de- and retrapping of the to observe the formation and decay of transient species in electron, or tunneling of the electron from a shallow t o a the 350-psec time window between fine-structure electron deep trap, is the probable mechanism for the solvation of pulses from the University of Toronto Linear Accelerator. et-. If this is the case, the amount of intermolecular hydroThe details of the SPR system are described elseThe Journal of Physical Chemistry, Vol. 79, No. 26, 1975

2836

I

W. J. Chase and J. W. Hunt Solvation of the Electron in I-Propmol at -12T

I-Propanol

I

-12T Midpoint

of pulse

‘Y

\

\ \

‘. Flgure 2. Typlcal plcosecond klnetlc traces of absorptlon vs. t h e In 1-propanol at -12OC. These traces were among those used to con-

struct the 30-psec spectrum of Flgure 1. The two vertlcal lines mark the middle of the pulse and the polnt at whlch the 30-psec spectrum of Flgure 1 was calculated. The absorptlon signal 1s nolsler at 1300 nm because the germanlum photodlode was used for thls trace, whereas the more sensltlve slllcon photodiode was used for the other traces.

here,^^^^^^^^^^^ and the effect that some of these details have upon kinetics is described in the Appendix. The width of the fine-structure electron pulse depends upon the alignment of the linear accelerator. This pulse width changes from run to run, and is checked each time by observing the formation of the absorption signal of eaq-, Chemicals. The DzO was supplied by Chalk River Nuclear Laboratories, Atomic Energy of Canada Limited. The DpO contained less than 1%HzO and less than 10-4 M of other impurities. Both DzO and once-distilled HzO were not degassed as it has been shown that dissolved oxygen has no effect on the yields or kineticsa23Ethanol was obtained from Consolidated Alcohol Limited or from U.S. Industrial Co. and all other alcohols were certified reagent grade from Fisher Scientific Co. The alcohols were not purified but were bubbled with argon gas (prepurified grade) from Canadian Anaesthetics Limited for at least 30 min before use. In alcohols, tests showed that dissolved air decreased the absorbance at 6 nsec by less than lo%, but left esul- yields and kinetics at picosecond times unaffected, Variable Temperature Irradiation Cell. To prevent a buildup of long-lived irradiation products the solutions must be continually flowed through the irradiation cell. At room temperature this presents no special problem and cells with optical path lefigths of 0.5,1, or 2 cm were used. A specially constructed cell allowed us to vary the temperature at which the liquid was irradiated. A 1-cm irradiation cell was suspended inside an aluminum housing which had windows of Suprasil quartz to allow the electron beam and the analyzing light to reach the cell. A vacuum provided thermal insulatiofi between the aluminum housing and the cell. The flowing liquid is first cooled by passage through a copper coil immersed in a bath of ethanol-dry ice mixture, and then heated by passage through an externally heated length of copper tubing. The temperature of the liquid can The Journal of Physical Chemistry, Vol. 70, No. 26, 1075

be varied from -65 to 45OC by this method, and by leaving the dry ice out of the cooling bath, temperatures of at least 6OoC can be reached. The temperature of the flowing liquid was sensed at the outlet of the cell by a temperature probe and converted to a voltage by a bridge circuit (Rosemount Engineering Co., platinum resistance temperature sensor, Model 146 MA, and linear bridge, Model 4142). The estimated error in the temperature measurement is less than l0C, although it should be noted that the electron beam heats the liquid by 3 or 4OC. All the temperatures quoted in this paper are those recorded during irradiation. Signal Detection. In these experiments, solid state photodetectors were used to detect the picosecond absorption signals. In the wavelength range from 300 to 1100 nm a silicon photodiode (United Detector Technology Inc., PIN-10) was used. This detector had a response time of 6 nsec into a 6 0 4 load and a maximum quantum efficiency of -90% at a wavelength of 900 nm. Observations have also been carried out at wavelengths as long as 1300 nm using a planar pussiuated germanium photodiode (RCA Limited, Montreal). This detector had a response time of 10 nsec and a quantum efficiency of about 30% at 1300 nm. The time resolution of the SPR system is not degraded by using this somewhat slower photodetector.

Results (1) Spectra. ( a ) HzO. As shown by Lam and Hun@ the picosecond and microsecond absorption spectra of eaq- are very similar. The absorption signal is formed in less than 10 psec and does not I “J or decay in 350 psec.23There is no indication of a fast decay in the infrared which could be attributed to et-. ( b )Alcohols. Unlike H20,the absorption spectra in alcohols change markedly with time. Baxendale and Wardmanll observed nanosecond spectral shifts in cold alcohols and we find similar spectral changes in alcohols on a picosecond timescale. A broad absorption band rising into infrared wavelengths is formed very rapidly; Basendale and Wardman observed that it was completely formed by the

Solvation Time of the Electron in Polar Liquids

2837

end of their 5-nsec pulse and we find that it is formed within the 18-psec response time of the SPR system. We attribute the infrared portion of the absorption band to electrons trapped in shallow potential wells. We have labeled these electrons trapped electrons, et-, to differentiate them from esol-, a fully relaxed electron which is presumably in a deeper potential well than et-. As et- becomes solvated

the shape of the absorption spectrum evolves with time, as shown in Figure 1 for 1-propanol at -12OC (261 K). The 30-psec spectrum changes to that of esol-, as shown by the 6-nsec spectrum. The 6-nsec spectrum peaks at 600 nm, compared to the value of 610 nm recorded by Dixon and Lopata30 at the same temperature. Our 6-nsec spectrum also has a shape similar to that measured for esol- at room temperature.22The spectra of Figure 1 are typical of those found for the other normal alcohols that we studied. Some of the picosecond kinetic absorption traces used to construct the 30-psec spectrum of Figure 1 are shown in Figure 2. A t all wavelengths a portion of the absorption signal (as indicated by the 30-psec spectrum of Figure 1) is formed rapidly within the fine structure pulse and then either grows or decays, depending upon the wavelength. A t short wavelengths (lo50 nm) the rapidly formed signal decays in 350 psec. At 1300 nm in Figure 2 the absorption signal decays to its baseline and there is no indication of a contribution from esol-. At 1050 nm the signal does not decay completely in 350 psec and has a small component apparently due to esol-. At intermediate wavelengths the signals form quickly and then decay slowly. At 900 nm in Figure 2, the signal is formed almost within the width of the fine structure pulse and then decays slowly in 350 psec. The fast formation followed by a slower decay agree with the results of Hase et al.7 and Baxendale and Wardman.ll These complex kinetics are probably due to a shift in the spectrum of et- as it relaxes into a deeper trap. A detailed analysis of the kinetics is difficult for wavelengths between 800 and 1000 nm. (2) Kinetics. The raw kinetic data are in the form of traces of percentage absorption vs. time, from which the absorbance, A@), can be calculated. From traces where A(t) returns to its baseline it was found that A(t) follows first-order kinetics at 1050 and 1300 nm A(t) = A(0) exp(-ktt)

(1)

and it was assumed that all of the solvation kinetics were first order a t these wavelengths. The growth of the absorption signal due to esol- at wavelengths shorter than -800 nm was also first order, as determined from traces where the signal is completely formed in 350 psec. For wavelengths less than -800 nm, A(t) is conveniently given by

A ( t ) = A(m)((l- a) + a [ l

- exp(-kso~t)])

(11)

where A ( - ) is the absorbance at infinite time (assuming no other reactions occur), (1- a)is the fraction of the absorbance which is formed during the fine structure pulse of the accelerator, and a! is the fraction of the absorbance which grows in with a rate constant, ksol. Rate constants, kt and ksol, were calculated from semilog plots of absorbance vs. time. For these plots the tilt of the

Formation of Solvated Electrons

700 nrn cm. cell Temperature = 26OC

'12

Figure 3. Experimental data and calculated curves for the formation of esol-. For both H20 and I-propanol the fine structure pulse shape was Gaussian with a u of 4.4 psec (see Appendix). For H20 the formation time used was ~~~1 = 0 psec. For 1-propanol the absorption buildup followed eq It with T~~~= 28 psec and a = 0.2.

baseline was determined from kinetic traces where the absorbance returned to its baseline, but the position of the baseline was shifted until the semilog plot was a straight line. The accuracy of individual rate constants determined from these plots was usually better than 20%, and in no case worse than 35%. , For lifetimes, T = l / k , shorter than about 20-psec kinetics could be altered by the 18-psec response time of the SPR system. For all traces with T < 20 psec and selected other traces a convolution technique (see Appendix) which accounted for this response time was used to calculate curves which could be compared to the experimental data, thus ensuring that the lifetime obtained from the semilog plot was correct. ( a ) Formation Time of esol- in H 2 0 and D20. The formation time, T,,L = l/ksol, of eaq- is known to be very rapid,13i23p27and in order to reduce the response time of the SPR system to a minimum, an irradiation cell with a path length of only 0.5 cm was used. As well, a narrow time range was studied so that the absorption buildup could be more closely examined. According to Bronskill et al.23 and Aldrich et a1.,18 the shape of the absorption signal is the convolution of the shape of the fine structure pulse, f ( t ) ,the desynchronization, d(t), between the electron beam and the slower light beam, and the function assumed for the concentration of the absorbing species, c(t). Details of this calculation are found in the Appendix. Beer's law relates c(t) to the absorbance, as given by eq 11. The fine structure electron and Cerenkov light pulses were assumed to be Gaussian in shape and are convoluted together to give f(t), itself a Gaussian.32 1 f(t) = -exp ( -t 2/4 u2) 2uvG in which u is the width of a fine-structure electron pulse. The formation of eeol- was observed a t 600 and 700 nm in HzO and a t 600 nm in DzO. A comparison between predicted curves and experimental data (see Figure 3) showed the following: (1) the fine structure pulse shape, f(t), is Gaussian or nearly Gaussian in general agreement with the The Journal of Physical Ctmrnhtty, Vol. 79, No. 26, 1975

2838

W. J. Chase and J. W. Hunt I14

a

Arrhenius Plot

20

order a t all wavelengths. They, are listed elsewhere; see the paragraph a t the end of the text regarding supplementary material. A t any given temperature ksol has the same value within experimental error at short wavelengths (400-800 nm) as kt has at long wavelengths (1300 nm). At intermediate wavelengths (900-1050 nm) k t is as much as 100%slower than the rate constants found in the other part of the spectrum, depending upon the alcohol studied and the temperature. We can accurately determine rate constants for decays of et- whose lifetimes are longer than about 20 psec since the response time of the SPR system with a 1-cm irradiation cell is 18 psec. Because there is such excellent agreement between the theoretical and observed formation of esol- in Figure 3 we can use the shape of the fine-structure pulse to calculate a curve for the decay of et-. By matching the theoretical curves to the experimental data we can obtain lifetimes for the decay of et- as short as 10 psec with reasonable accuracy. The kinetics of the decay of et- a t 1300 nm are the least ambiguous because the absorption signal always decays to the zero absorption level as there is no esol- absorption signal. For this reason the kinetics of the decay of et- a t 1300 nm are discussed first. (i) 1300 nm. An Arrhenius plot of the rate, kt, of disappearance of et- is shown in Figure 4a,b for methanol, ethanol, and 1-propanol over a temperature range from -65 to 6OOC. The observed decay rates change by a factor ranging from 5 to 10 over this temperature range. At 20°C the values of 7(et-)1300 nm are 10.7 f 1, 23 f 2, and 34 i 3 psec in methanol, ethanol, and 1-propanol, respectively. In Figure 4a the line for methanol is a least-squares fit of the data to an Arrhenius expression

1

10.c r

1.0

l-P,oPo"ol

05

5L

3033

0.004

0.0045

0005

~/rc'K-l;

iE

k

0003

00035

0004

k

0 w45

0035

IK' '1

Flgure 4. (a) Arrhenius plot of the rate constant for the decay of etmeasured at 1300 nm in methanol (0)and 1-propanol (A).The solid line for methanol is a least-squares fit of the data to the Arrhenius expression (eq IV). The solid line for I-propanol is a least-squares fit of the data to the Cole-Davidson expression (eq V) with TO = 73.5 K. (b) Arrhenius plot of the rate constant for the decay of et- measured at 1300 nm in ethanol (#). The dashed line (- - -) is a leastsquares fit to the Arrhenius expression (eq W).The solid line (-) is a least-squares fit to the Cole-Davidson expression (eq V) with TO = 73.5 K. The parameters of the curves in a and b are listed in the supplementary material. results of Mavrogenes et al.;31 (2) the formation time, r,,~, of esol- in HzO and DzO is between 0 and 3 psec. Figure 3 shows that the calculated curve for the absorption buildup (using u = 4.4 psec and rsol = 0 psec) agrees extremely well with the experimental data. It should be noted, however, that by using u = 4.0 psec, rsol = 2.0 psec and a = 1, a curve nearly identical in shape with, but displaced to the right of, the curve in Figure 3 is generated. Unfortunately we do not have an accurate, absolute time marker and thus we cannot distinguish between these two cases. Finally, if a were to have a small value of 0.2, there would be a good fit between the calculated curves and the experimental data for values of 7,ol as large as 10 psec. From our picosecond studies of HzO, there is no evidence for a short-lived absorption in the infrared due to the decay of et-; thus we consider it most unlikely that 7,01 could be more than a few picoseconds. ( b ) Alcohols. The rate constants for the decay of et- and for the formation of esol-, k t and ksol. respectively, are first The Journal of Physical Chemktry, Vol. 79, No. 26, 1975

(IV) while for 1-propanol the curve is a fit to a modification of eq IV due to Cole and D a v i d ~ o n . ~ ~ = A exp(-E,/RT)

k = A exp(-E,/R(T

- To))

(V) In this equation, To represents a glass transition temperature. In agreement with dielectric relaxation a value of 73.5 K was chosen for Figure 4a. The least-squares fit for 1-propanol includes recent new data from B a ~ e n d a l eIn .~~ Figure 4b the dashed and solid lines are least-squares fits to Arrhenius and Cole-Davidson expressions, respectively. (ii) 1050 nm. At 1050 nm the decay of the absorption signal is similar to that seen at 1300 nm except that a component due to esol- may remain, depending upon the temperature and the particular liquid used. After accounting for this the data can be analyzed exactly like the data at 1300 nm. As shown in the Arrhenius plot of Figure 5a,b the rate constants measured at 1050 nm in methanol, ethanol, 1propanol, and 1-butanol are somewhat slower than those measured at 1300 nm. The analysis was done carefully and we feel that the differences are real and probably due to complex spectral shifts as et- becomes (iii) 600 and 700 nm.The formation of esol- was studied at 600 and 700 nm. At 700 nm definitive values of rS01 could not usually be obtained because the fraction of the signal, a, of the slow formation of esol- was small (-0.21, but all of the data is consistent with eq 11. In Figure 3 the absorption signal of esol- in 1-propanol is compared to the calculated curve using a = 0.2 and 7~01= 28 psec. It should be noted that a wide range of values for 7~01(10-50 psec) is consistent with the experimental data. A more accurate assessment of 7~01was obtained a t 600

Solvation Time of the Electron in Polar Liquids

2839

nm a t a variety of temperatures. At 600 nm a is larger and at cold temperatures a definite break can be seen between the fast (1 - a ) and slow ( a ) components of the esol- absorption signal. The rate constant, ksol,was calculated from the slow formation component and is plotted in Figure 5a,b for methanol, ethanol, 1-propanol, and 1-butanol. The curves are from the 1300 nm data and those of Figure 4a,b; there is a good agreement between the two sets of data a t the different wavelengths. ( c )Formation Time of esol- in Ethylene Glycol. Because the signals were noisy accurate estimation of ~~~l could not be made by observing the forqation of esol- at 600 nm at room temperature. However, a signal due to et- in ethylene glycol has been observed at 1000 nm by Lam and Hunt.14 This signal decays at a rate faster than that for the decay of et- in methanol and it is not possible to obtain an accurate rate constant even using a convolution technique. However, we observe that at cold temperatures the relative absorptions of et- in the different alcohols are the same to within -30%; these absorption yields are also nearly independent of temperature provided a correction is made for the decay of the absorption during the fine structure pulse. Thus if we assume that the absorption yields are the same in methanol and ethylene glycol we estimate that et- decays twice as fast in ethylene glycol as in methanol. Our crude estimate for the formation time of esol- in ethylene glycol is ~ , , 1 = 5 f 3 psec.

0003

00035

0004

0 0045

0 005

VT ("K"

1'16

{"A

Discussion We have been studying electron solvation in highly polar 98 liquids that form large numbers of intermolecular H bonds. As a result, these liquids are highly associated and it has 96 been suggested that pure ethanol-d1 exists primarily as a 88 94 solution of tetramers.36 From dielectric relaxation studies the activation energy, E,, for H bond breakage ranges from 86 0003 0035 'IT 0004 I'K ') 0045 0005 92 3.5 kcal/mol in methanoP7 to 5.6 kcal/mol in l - b u t a n ~ l . ~ ~ From our data (Figures 4 and 5) the apparent activation Figure 5. Arrhenius plot of the rate constants for the decay of etenergy for solvation ranges from 2 to 3.5 kcal/mol near (solid symbols, measured at 1050 nm) and for the formation of ewlroom temperature. Thus the solvation process cannot be (open symbols, measured at 600 nm). The curves are from Figure 4a,b. explained by H-bond breakage alone, nor by viscosity alone, as solvation occurs in alcoholic glasses a t 77 K9J0J2 where the viscosity is very large. Certainly, electron solvation cannot be explained by a single process, as an Arrhentation produce most of the chemical changes. In addition, it ius plot of our data and that of B a ~ e n d a l eshows ~ ~ a dishas been argued for low LET radiation where the energy is tinct curvature. The dielectric relaxation for molecular deposited in "spurs" or "blobs", that the temperature rise rotation should perhaps show a good correlation with ~ ~ ~ would 1 , neither be high enough nor last long enough to sigwhich also involves molecular reorientation. However, nificantly alter the reaction r a t e ~ . * ~However, ,~l since the there is a fundamental difference in the two processes; in solvation process has a rate constant of 1O'O to loll sec-' dielectric relaxation the reorienting field is external to the around room temperatures in the alcohols and takes place medium while in electron solvation it is internal. Nonetheentirely in the spur, it is necessary to reconsider whether a less, the dielectric relaxation time for molecular reorientatemperature jump could affect the rate of solvation. tion corresponds rather well to the time required for elecFollowing the theoretical analysis of Magee et a l j Oand tron solvation. Mozumder?l we can obtain an estimate of the temperature As well a problem peculiar to ionizing radiation will be jump. If we assume that the ionizing radiation deposits its considered. The ions and electrons are not distributed ranenergy in a spherical spur of radius ro, the excess temperadomly, but ihstead they are clustered along the tracks of ture T,,(t) relaxes in time, t , as given by40p4l the charged particles (spurs). A certain fraction of the enerT , , ( t ) = A T ( ( 1 + 4 D t / r 0 * ) - ~exp[-r2/(ro2 /~ + 4Dt)l) gy expended is not expressed in ionization, but instead as (VI) local "hot spots" within the spurs. The effect of this temperature jump on the solvation processes will be estimated. where AT is the maximum temperature rise at the center of A. Temperature Jump. It was once thought that "pointthis spur and r is the distance from the center of the spur. heat" or local heating effects could account for some of the The quantity D is the thermal diffusivity, given by chemical changes induced by radiation,39 but subsequent radiation chemical studies proved that ionization and exciD = KIpC, (VII)

:

I

:

The Journal of Physical Chemistry, Vol. 79, No. 26, 1975

2840

W. J. Chase and J. W. Hunt

0.003

a004

0.005 ‘/T

a m 1180 D

(OK“)

-

Figure 6. The effects of a time dependent temperature jump, Tex(t), in altering the effective rate constant, k = l / T e f f of etesol-. in 1-propanol at the center of the spur ( r = 0 in eq VI). The effective lifetime, Teff, is calculated from equation (X). The solid curve (-) is calculated using T,,(t) = 0 (no temperature jump) and is the same as the curves in Figures 4a and 5a. The curve (-) is the effect of Tex(t),given by eq VI, when the maximum temperature jump is A T = 56OC.The curve (- -) for Tex(t)shows the effect of neglecting thermal diffusion; this is an upper limit to the effect of the temperature jump.

--

-

where K is the thermal conductivity, p is the density, and C, the specific heat at constant volume. The maximum temperature rise at the center of the spur is therefore given by AT = Edep/~3’2r03pCv (VIII) where Edep is the energy deposited in the spur that becomes heat. The time required for the excess temperature to drop to one-half of its original value is given by the solution of

+

(1 4 0 t / r 0 ~ ) - = ~ /1/2 ~

(IX)

which for 1-propanol (using K = 1.409 X loe3 J sec-l cm-l K-l) is 15 psec. Depending upon the temperature sensitivity (apparent activation energy) of the reaction rates, a temperature jump could alter the rate of reactions, such as electron solvation, having reaction lifetimes from 10 to 100 psec. Let us now consider the effect of the temperature jump on reaction 1. The differential equation for the rate of disappearance of et- is d[et-]/dt = +[et-]

(XI

where k , instead of being a constant as is usually the case, is now time dependent. If k shows an altered Arrhenius type temperature dependence of the Cole-Davidson type (eq V), the instantaneous value of k is given by combining eq V and VI

k ( t ) = A exp[-E/R(T - To + Tex(t))]

(XI)

where T is the ambient temperature. A computer program was written to solve eq X with a The Journal of Physical Chemistry, Vol. 79, No. 26, 1975

time dependent rate constant given by eq XI. In order to compare reaction rates, an effective rate constant for the solution of eq X can be defined as keff = l/Teff,where Teff is the time required for [et-] to be reduced to l/e of its original value. Unfortunately, accurate values for the spur parameters are not available because of the apparent discrepancies between the theoretical and observed decay of eaq- in s p ~ r s . ~ Mozumder41 l-~~ has made an estimate of E d e p = 30 eV and ro = 20 A in HzO. Using these values in eq VIII, for 1-propanol the maximum temperature at the center of the spur is AT = 56OC, which would increase k considerably. This is illustrated in Figure 6, an Arrhenius plot of the effect of a 56OC temperature jump on the decay of et- in 1-propanol. The lower curve is the one for l-propanol used in Figures 4a and 5a. The middle curve is the effective rate constant, keff, for AT = 56OC. At room temperature keff is -50% greater than the rate constant for no temperature jump. Note that this suggests the formation time of esol- produced by an ionizing beam could be considerably faster than for esol- produced a photoionization process.27The upper curve is a calculation using Tex(t) = A T = 56OC and is an upper limit to the effect of the temperature jump, since Te,(t) I A T for all time and space. It should be pointed out that the spur theories do not agree with the observed spur decays+s therefore the corrections to ksol by the temperature jump might be incorrect. For example, Kupperman4*has recently suggested that the spatial distribution of eaq- in the spur might be wrong. He estimates that the spur contains 6.7 ion pairs and has an initial radius, ro, of 60 A. Using this data, Edep N 95 eV, but the temperature jump is only -7OC; this temperature jump is so small that no corrections would be needed. Because of the uncertainty of the possible temperature jump correction, no such correction was applied to the observed rate constants in this paper. B. The Solvation Time of the Electron in Liquid Alcohols. ( a ) Dielectric Relaxation. As background before considering the solvation process in detail it is useful to discuss the theory of dielectric relaxation. On a molecular level these two phenomena are similar and it might be expected that Tsol should be correlated with one of the characteristic dielectric relaxation times. When an electric field is applied to a dielectric medium, the resulting polarization is proportional to the dielectric constant, t. If the electric field is changing rapidly, the polarization will lag behind the electric field and the dielectric constant now is a complex quantity. The imaginary part of e is the dielectric loss factor, e”. Frequency regions where the dielectric losses are nonzero are called dispersion regions. Dielectric relaxation measurements have shown that there are three distinct dispersion regions in the normal alcohols for l -propanol and its higher homologue^.^^^^^ Only two distinct dispersion regions have been observed in methanol and ethan01.4~9~7 For each dispersion region, x , the real part of the dielectric constant, e’, can be described by t’ = emX - (tox - tmx)/(l+ W 2 T x 2 ) (XI11 where T~ is the dielectric relaxation time in the region x . The dielectric constants, eox and cmX, are the limiting low and high frequency dielectric constants in each region. The longest dielectric relaxation time, 71, is the time required for the breaking of intermolecular H bonds followed by reorientation of the liberated molecules. Most of the

Solvation Time of the Electron in Polar Liquids

2841

-

I

5 60

-

vs Number of Carbon Atoms

Cole-Dovidson p o t I- Propono

e l300nm

Go, 0 600 nm 514 nm

50 -

u P

0

72

o

40-

e

20

-

IO

-

e

X

0

0

x

k

X

0

30-

x

X

x

x

P A e

A i‘,,, .Chase and Hunt A

changes in the dielectric constant occur in the first dispersion region which implies that most of the molecules must break intermolecular H bonds before they can rotate. The intermediate relaxation time, 72, is the time required to reorient a monomer in the liquid, The shortest time, 73, is the reorientation time of OH dipoles. When an electron is injected into a medium, the charge of the electron is thought to speed up the process of dielectric relaxation48149and a dielectric relaxation time, r’, at constant charge (the “constant charge approximation”) is given by

(XIII) where r x , e,, and eox have been previously defined. The possible effect of the “constant charge approximation” is particularly important for 71; for the other dispersion regions, emx/Bo, changes 7, by less than a factor of 2. Because the alcohols are strongly associative it might be expected that 7,1 and 71’should be correlated. ( b ) Comparison of Dielectric Times and ~ , , i . The dielectric relaxation times, 71,72, 73, and TI’, are compared to r , , ~ in Table I for HzO and the first four primary alcohols. The

T,,I

- Baxendola sini

values of 71 are much larger, and 73 much smaller, than T,,I, while rz is similar to 7,1. As well, the modified dielectric relaxation time 71’ of Schiller48 and Mozumde+* (see eq XIII) corresponds closely to the observed rsol for methanol and ethanol. However, the ratio, T I ’ / T ~ ~increases I as the number of carbon atoms in the alcohol also ,increases. Therefore, it appears that r8,i is not closely correlated to 71’; for example, 71’/7‘801 in 1-butanol is 3.4. The close relationship between rz and r,,l is emphasized in Figure 7, where 7s01and 72 are compared for a number of primary alcohols. The values of 7801 and r2 agree to within a factor of 1.6. The correlation between ~ , , i and 72 at different temperatures is shown in Figure 8, in which the values of r8,,1in our work and the work of B a ~ e n d a l eare ~ ~plotted in a Cole-Davidson pl0t.3~As well, dielectric relaxation t i m e ~ ~ 3 ~71, 3 ~7J3 ’and ~ rl’, and the viscosity,61Bsz 7, are compared to rn01.This plot indicates that T,,I corresponds closely to r~ over the 215 K temperature rahge. The modified time, TI’, is considerably slower than real, particularly at the lower temperatures. One should note as well that viscosity data are parallel to the ~~~i data. Plots similar to Figure 8 are obtained for methanol, ethanol, and 1-butanol; there The Journal of Physloel Chemistry, Vol, 70, No. 26, 1075

W. J. Chase and J. W. Hunt

2042

are similar correlations between 72 and 7,1 over a wide range of temperatures. The correspondence between 7 2 and 7,1 rather than 71’ and r,,~implies that the large amount of intermolecular H bonding present in these alcohols does not play a large part in the process of electron solvation. It is possible that this correspondence is just a coincidence and that the theory of the constant charge modification to 71 is correct in principle but not in its detailed application to the dynamics of solvation. It is more probable, however, that the intermolecular H bonds are broken very rapidly by the strong electric field of the electron, and that the rotation of the molecules limits the rates of the solvation processes. If this is true, the solvation time would be 7 2 (or possibly 72’) as the alcohol molecules reorient around the electron. Wave Packet Model. In this model by Schiller and a moving electron polarizes the liquid to form a potential well in which the electron becomes self-trapped. The electron is described by a “roof” shaped wave packet. Upper, tmb, and lower, tma, estimates for r,,~are combined to give the intermediate value (tmatmb)”’, as listed in Table I. It can be seen that 17,, is always larger than (tm,tmb)1/2but considering the approximations made in the model the agreement is rather good. In Table I of ref 53 it can be seen that T , ~ Iand (tmatmb)1’2also agree to within a factor of 2-4 at colder temperatures except for methanol. Overall, this model predicts ~~~l with reasonable accuracy, The Fueki-Feng-Kevan Semicontinuum Model. In the Fueki-Feng-Kevan semicontinuum the medium surrounding the electron is divided into a shell of N ( N = 4 or 6) symmetrically oriented solvent molecules, and the molecules surrounding the first shell are treated as a continuous dielectric medium. During solvation, the electron reorients the dipoles of the molecules in the first shell by a short-range charge dipole attractive potential and polarizes the surrounding medium by a long-range polarization potential. An estimate of 7,1 was made by calculating the relaxation time, T ~ for , orientation of the dipoles in the first shell using the classical formulation of D e b ~ e The . ~ ~rate of dipole orientation is limited by the microscopic viscosity, 9 , of the alcohol; in the model it is then assumed that can be approximated by the microscopic viscosity. The dipoles are originally oriented at an angle, 00, to the central electron. The times, T ~ for , dipole orientation are given in Table I for 00 = B O O . For methanol, ethanol, and 1-propanol T~ is faster than 7,1 by factors of 5, 3, and 2, respectively. It should be noted that the two theoretical estimates, (tm,tmb)1/2and T ~ are , in very good agreement. In Figures 1-3 of ref 54 the temperature dependence of T~ is compared to experimental data. The curves for 1-propanol fall on most of the data points and reproduce the form of the curvature shown in Figure 4a of this paper. The experimental data for methanol and ethanol do not fall on the curves for T ~ In . the case of methanol T~ is much shorter than the experimental values of ~ ~ In~ general 1 . the curves for T~ predict the form of the temperature dependence observed for ~ , , l but cannot as yet always predict the values of ~ ~ Modifications ~ 1 . taking into account the change of cavity radius around the electron as the dipoles orient, and polarization of the surrounding continuum do not appreciably alter the predictions of the semicontinuum It would seem to us that it is necessary to take into account dipole-dipole interactions before a better agreement between theory and experiment can be reached. The Journal of Physical Chemistry, Val. 79, No. 26, 1975

Conclusions (1) The excellent correlation between the rates of decay of et- and of formation of esol- is strong evidence that etreally is the immediate precursor of esol-. (2) The solvation time of the electron is not correlated with the dielectric relaxation times 71, for H bond breakage, and 73, for OH dipole rotation. When 7 1 is modified to account for the charge of the electron, the modified time, TI’, agrees with 1,7, for methanol and ethanol. However, 71’ is considerably longer than T,,I for the larger alcohols. This implies that the rate of electron solvation is not limited by the rate of H bond breakage. (3) The dielectric relaxation time for molecular rotation, 72, shows a very good correspondence to T,,I over a wide range of temperatures for the alcohols. Thus molecular reorientation, rather than thermal de- and retrapping or trap-to-trap tunneling of the electron, appears to be the mechanism of electron solvation.

Acknowledgments. The authors thank A. Worthington, D. Bazett-Jones, and M. O’Loughlin for their help in doing the experiments and H. Yee as a summer student for his computer programming. We acknowledge some very useful comments during the preparation of this paper from Drs. G. A. Kenney-Wallace and J. H. Baxendale. Dr. A. D. Singh, Whiteshell Nuclear Research Establishment, Atomic Energy of Canada Limited, arranged the loan of the D20. We also acknowledge the continuing help of Mr. T. Horrigan and the staff of the Linac Laboratory of the University of Toronto. The experimental work was supported by the Medical Research Council of Canada and the National Cancer Institute of Canada. Financial support for the operation of the Linear Accelerator at the University of Toronto was provided by the National Research Council of Canada. Appendix Because the kinetics that were studied were very rapid it was necessary to be able to compare our experimental results with theoretical calculations which take into account the complexities of the SPR system. To do this computer programs, similar in principle to those previously used,l* were written to generate theoretical curves of optical absorption vs. time which could be directly compared to our picosecond kinetic traces. In these calculations the total response function (the fine structure pulse shape, s ( t ) )of the SPR system is generated from the individual system response functions and then convoluted57 with the radiation chemistry that occurs in the cell, the result being the radiation chemical response function of the SPR system to a single fine structure electron pulse (the fine structure pulse chemical response function, r( t ) ) .Because the electron macropulse consists of a train of fine structure pulses, the overall absorption is the sum of the absorptions of the fine structure pulses, each of which is a different age. To generate the overall response of the system the individual fine structure pulses response functions must, except under certain circumstances (see below), be appropriately summed. Several factors contribute to the fine structure pulse shape but some of these are not important. The dose delivered to the sample was previously assumed to vary linearly from 1.2 to 0.8 (in relative units) from the front to the back of the It was found that this dose distribution was in-

2843

Solvation Time of the Electron in Polar Liquids

distinguishable from a uniform dose distribution and therefore the latter was routinely used. The electronic response time, s , ~ ~depends ~, upon the scanning rate of the light delay mirrors and the smoothing circuit time constant.lsvZ3 Generally, 7elec is less than the equivalent of 2 psec and for particularly sensitive experiments, such as the formation of eaq-, 7&c was reduced to the completely negligible value of 0.3 psec. The important contributions to the fine structure pulse shape are as follows. (1)The electron distribution is assumed to be a Gaussian of the form exp(-t2/2u2). (2) The Cerenkov analyzing light pulses are produced by the electron pulses and thus have the same shape. When the two Gaussians are convoluted the u2 add and the combined response function is also Gaussian. We shall refer to this function as the fine structure pulse function 1 f(t) = -exp(-t 2 / 4 0 2 ) 2 6 G

as shown in Figure 9. (3) A photon takes T’ psec longer than an electron to traverse the sample. This “desynchronization time” is

T’ = (n - l)(l/c)

(XV)

where n is the index of refraction of the sample, 1 the path length of the sample, and c the speed of light. For water in a 2-cm cell T‘ = 22 psec. Thus an instantaneous event is spread out to one over T‘ psec. The desynchronization response function is: d(t)=l =O

O